aya-23-35B-IMat-GGUF

Llama.cpp imatrix quantization of CohereForAI/aya-23-35B

Original Model: CohereForAI/aya-23-35B
Original dtype: FP16 (float16)
Quantized by: llama.cpp b2998
IMatrix dataset: here

Files

IMatrix

Status: ✅ Available
Link: here

Common Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
aya-23-35B.Q8_0/*	Q8_0	37.18GB	✅ Available	⚪ No	✂ Yes
aya-23-35B.Q6_K.gguf	Q6_K	28.71GB	✅ Available	⚪ No	📦 No
aya-23-35B.Q4_K.gguf	Q4_K	21.53GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.Q3_K.gguf	Q3_K	17.62GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.Q2_K.gguf	Q2_K	13.82GB	✅ Available	🟢 Yes	📦 No

All Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
aya-23-35B.FP16/*	F16	69.97GB	✅ Available	⚪ No	✂ Yes
aya-23-35B.Q5_K.gguf	Q5_K	25.01GB	✅ Available	⚪ No	📦 No
aya-23-35B.Q5_K_S.gguf	Q5_K_S	24.34GB	✅ Available	⚪ No	📦 No
aya-23-35B.Q4_K_S.gguf	Q4_K_S	20.38GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.Q3_K_L.gguf	Q3_K_L	19.15GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.Q3_K_S.gguf	Q3_K_S	15.86GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.Q2_K_S.gguf	Q2_K_S	12.74GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ4_NL.gguf	IQ4_NL	20.23GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ4_XS.gguf	IQ4_XS	19.20GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ3_M.gguf	IQ3_M	16.70GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ3_S.gguf	IQ3_S	15.86GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ3_XS.gguf	IQ3_XS	15.09GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ3_XXS.gguf	IQ3_XXS	13.83GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ2_M.gguf	IQ2_M	12.68GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ2_S.gguf	IQ2_S	11.84GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ2_XS.gguf	IQ2_XS	11.10GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ2_XXS.gguf	IQ2_XXS	10.19GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ1_M.gguf	IQ1_M	9.15GB	✅ Available	🟢 Yes	📦 No
aya-23-35B.IQ1_S.gguf	IQ1_S	8.52GB	✅ Available	🟢 Yes	📦 No

Downloading using huggingface-cli

First, make sure you have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Then, you can target the specific file you want:

huggingface-cli download legraphista/aya-23-35B-IMat-GGUF --include "aya-23-35B.Q8_0.gguf" --local-dir ./

If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run:

huggingface-cli download legraphista/aya-23-35B-IMat-GGUF --include "aya-23-35B.Q8_0/*" --local-dir aya-23-35B.Q8_0
# see FAQ for merging GGUF's

FAQ

Why is the IMatrix not applied everywhere?

According to this investigation, it appears that lower quantizations are the only ones that benefit from the imatrix input (as per hellaswag results).

How do I merge a split GGUF?

1. Make sure you have gguf-split available
   • To get hold of gguf-split, navigate to https://github.com/ggerganov/llama.cpp/releases
   • Download the appropriate zip for your system from the latest release
   • Unzip the archive and you should be able to find gguf-split
2. Locate your GGUF chunks folder (ex: aya-23-35B.Q8_0)
3. Run gguf-split --merge aya-23-35B.Q8_0/aya-23-35B.Q8_0-00001-of-XXXXX.gguf aya-23-35B.Q8_0.gguf
   • Make sure to point gguf-split to the first chunk of the split.

---

Got a suggestion? Ping me @legraphista!